Modeling and Implementation of Intelligent Optimization Scheduling System for Enterprise Carbon Emission Reduction Based on Reinforcement Learning

In this paper, we propose an intelligent scheduling system based on deep reinforcement learning for carbon emission reduction in enterprises. By analysing equipment operation and load characteristics, we establish a multi-objective reward function that considers carbon emissions, productivity and operating costs. The system adopts a distributed architecture with multiple data collection layers for real-time monitoring and optimisation. Experiments demonstrate that our approach significantly reduces carbon emissions and energy consumption while maintaining production requirements and improving efficiency. The performance metrics exceed traditional scheduling methods and provide a viable solution for low-carbon transformation of manufacturing companies. Combining advanced machine learning techniques with industrial processes demonstrates great potential for optimising energy consumption patterns across the manufacturing industry. This research contributes to the growing body of knowledge on the practical application of AI in addressing the challenges of climate change while maintaining economic viability, providing a balanced approach to the competition between environmental sustainability and industrial productivity. In addition, the scalability of the proposed system allows it to be adapted to a variety of industrial environments, potentially accelerating the transition to carbon-neutral manufacturing across industries.